SwiftBot Autonomous Delivery Robot
The main problem that this project seeks to address is the safe and timely delivery of small payloads such as food items, logistical supplies and equipment, and documents in indoor and closed campus spaces. The project involves the implementation of an automated delivery service based
2025-06-28 16:36:13 - Adil Khan
SwiftBot Autonomous Delivery Robot
Project Area of Specialization RoboticsProject SummaryThe main problem that this project seeks to address is the safe and timely delivery of small payloads such as food items, logistical supplies and equipment, and documents in indoor and closed campus spaces.
The project involves the implementation of an automated delivery service based on a wheeled mobile robot and a mobile app that allow users get items delivered within indoor and closed campus spaces such as universities, offices and hospitals. The robot is based on the ROS (Robot Operating System) meta operating system and embedded system platforms which communicate with a central server for localization, mapping and obstacle avoidance computations as well as for coordinating deliveries. The goal is to automate the delivery process for physical items in closed campus spaces for enhanced convenience and reducing chances of error in physical handling and delivery of items.
Project ObjectivesThe need for secure and “swift” delivery has always been a problem within many areas of our society and there are also issues of moving items within an organization that can hinder or delay tasks which must be completed within a fixed time frame. There exists a requirement for tackling the issue of deliveries within closed spaces in a safe and timely manner. This project proposes an autonomous unmanned ground vehicle that utilizes simultaneous localization and mapping (SLAM) capabilities to plan a path to a destination intended for delivery of the load. A mobile application integrated with a mapping service would be utilized by a user to mark a destination area or coordinate and using data harvested from an integration of sensors such as cameras utilizing object detection, LIDARs and ultrasonic sensors, the mapping and path planning process can be used to automate the delivery process between two points. Security features such as facial recognition or app-based control are proposed to be incorporated to secure the payload within the robot. Part of the motivation of the project stemmed from observing transportation needs on the Habib University campus from receiving copies of payment cheques and receipts from the Career Services Office to moving exam papers and lab manuals between different offices/departments when dealing with bulk loads and downed printing services. With the group’s interest in robotics and wish to make campus life easier for its denizens, the group thought about potential use cases for the proposed project and how it could be scaled for other needs which led it to proposing this idea.
In summary, the wider aim of this project is to indigenously develop a prototypal delivery robot service which would handle small payload deliveries such as food parcels, small lab materials and document stacks in confined environments such as office floors and campus spaces. The further vision with this project is scalability to handle deliveries for larger complexes and on multiple floors but for the purposes of demonstrability of the concept, we are limiting the project to single floors and indoor spaces. The robot would be operating in a university campus setting and users can primarily interact with the robot through a mobile application to manage what item to order, track the progress of a delivery and monitor other associated delivery stats. Based on the setting and primary users (i.e. university students, staff etc.), strict decorum and codes as in the case of more controlled environments like hospitals and industrial offices are not a major concern. Only any applicable university codes would need to be considered. It is desirable for the solution to not be too bulky and noisy so as not to disturb the peace of a university campus atmosphere and so that the robot is more easily integrated in a campus dominated by a sizeable human population.
Project Implementation MethodThe following develpment goals are required for the final system:
- Autonomous navigation of a delivery robot with 2D environment mapping capability.
- Path planning behavior towards a goal.
- Obstacle detection and avoidance.
- App service for coordinating and managing deliveries.
- Admin panel for monitoring robot and delivery status and remotely controlling the robot for mapping purposes as required. Status indicators may include observables such as battery level, delivery route information, some robot state estimates such as current heading and speed as desired.
- API controlled locking mechanism for physically securing the payload.
- Software based security measure for user verification and access to app and payload - a user facial recognition feature.
System Constraints and Design Considerations:
- Accuracy of map is based upon choice of sensor. Currently a LIDAR is used with a 10 Hz scanning frequency which produces around 4000 range samples per second and has a distance range of 6m. The map producing algorithm (e.g. Hector SLAM or Gmapping) also affects the fidelity of the map.
- The robot’s odometry is limited by the sensor choice and accuracy as well as the technique used for odometry e.g. the resolution of the encoder motor that is used for dead reckoning. The final accuracy value is based upon our final sensor choice. For the proof of concept, a Pololu 70:1 gear ratio 150 rpm quadrature encoder motor was used with a pulse per revolution count of 4480.
- The maximum payload size and weight that can be handled by the robot is dependent upon the design decisions taken about the robot’s body. We are targeting payloads of around 1 kg and the dimensions would be based upon the payload holding compartment on the robot.
- The type and rating of the battery would determine its operating time and range. We are currently targeting Lithium Polymer batteries due to reduced size and weight requirements.
-
The robot's mechanical design has to be robust enough to support payload weight, actuators and connected components so as to last for maximum operational cycles. The design should also possess a high degree of modularity for ease in changing components during maintenance instead of reattaching parts from scratch.
-
In a lot of literature surveys, we have found that similar solutions are often expensive for the adopting organizations thus limiting these systems to deployment in only a few locations; therefore, a requirement in minimizing costs is selection of hardware including sensors, embedded controllers, actuators and connecting components so that the project does not compromise on the accuracy, robustness and inherent quality of the delivery system while being able to produce an affordable solution for smaller organizations.
The motivation of the project initially stemmed from observing efficiency and convenience factors of transportation on the Habib University campus such as for food deliveries from the cafeteria and moving equipment between labs in a safe and rapid manner. After literature surveys, the transportation problem was also discovered in other scenarios such as in hospital settings where time-critical deliveries such as lab sample delivery, medical instrument delivery etc. are required. Existing solutions that focus on automating hospital logistics via robots include the Help Mate robot which was commissioned in the late 90s. Besides possessing a dated design, the cost of purchasing this robot lies around US $110, 000 and weekly rentals cost US $25,000 making this solution accessible to a select few large organizations. Additionally, the use of an automated delivery system also minimizes the risks of mishandling of senstiive equipment via human couriers and also relieves nursing staff in hospitals to focus more on patient care as opposed to mundane item transportation tasks. Our surveys also led us to discover other attempts at tackling the above mentioned problem e.g. Amazon’s Scout delivery robot and Starship’s robot delivery service working in the area of short range deliveries in particular settings like university campuses neighborhoods, and office spaces whose occupants serve to constitute some of the relevant stakeholders associated with the problem. However, these solutions are contextualized and optimized for the neighborhoods and spaces they are operating within. The market for service robotics is also a booming one. A recent robotics and AI research partnership between Sony and Carnegie Mellon University aimed at optimizing the food preparation and delivery process in confined environments is one among a number of examples which highlight the growing investment in service robotics and the potential areas of benefit. Another use case highlighting the efficiency of robots for deliveries in closed spaces is exhibited by the use of wheeled mobile robots in a Las Vegas hotel (The Renaissance) which delivers toiletries and sundry items to hotel guests, essentially serving as waiting staff. All in all, the market for service robotics and the multitude of use cases to benefit everyday life for people - from administrative personnel, patients and nursing staff in hospitals to students and office workers on campuses - presents a viable opportunity for working in the space of delivery robots for indoor spaces.
Technical Details of Final DeliverableSystem Architecture:
https://drive.google.com/file/d/17EVRLD7bD3Fnih2M9IYWGJUnwhRmrQkj/view?usp=sharing
Physical Architecture:
https://drive.google.com/file/d/1RfcRYIHZC5i9LyoigBjHxzuU6ogiHUVx/view?usp=sharing
EE aspects:
- Design of optimal robot perception system using sensor combination – LIDAR, cameras, ultrasonic sensors.
- Design and implementation of control system for locomotion – PID control for smooth movement, motor control circuitry etc.
- Communication module – between embedded systems and server – via consideration of specific protocols and wireless systems for communication (e.g. WiFi)
- Hardware construction: robot model designing and implementation – this involves CAD models, base design, sensor attachments, payload securing attachment, and other hardware attachment design and analysis as well as PCB design for circuitry and so on.
EE/CS overlapping aspects:
- 2D Space Mapping – A laser scanner (LIDAR) is utilized for detection of the surrounding as the modules commonly available have built in actuation mechanisms to rotate the sensor to scan the surrounding in a 2D plane in a power efficient manner with low sound disturbance. This is opposed to cameras or a Kinect sensor for which we would have to design the rotation mechanism and manipulate the input data to reflect mapping in a 2D plane. The sensor used here has a relatively high degree of accuracy up to a 6m range.
- SLAM Method and Path Planning – this involves creating a map of the surrounding while simultaneously localizing the robot within the environment being mapped. This would form the basis for the path planning algorithm for delivery to the destination. The ROS operating system has some built in support for SLAM such as Hector SLAM and Gmapping methods including filters for cleaning the input data that we can configure and optimize for our requirements.
- Obstacle detection and avoidance – this is required to navigate the environment without bumping into obstacles both static ones, like furniture and walls as well as dynamic ones such as humans. This is achieved by using the LIDAR’s laser scan to calculate distance and orientation with obstacles.
- Remote Operation Functionality – This would be intended for monitoring purposes (robot location, delivery status etc. for admin) and initially mapping an area manually if required.
CS aspects:
- Client – Server communication implementation (using PHP) for the robot and mobile application to send and receive data and computations.
- App development for the delivery system and web based admin panel for overarching monitoring and control of operations.
- Database implementation (using MySQL) for holding records and managing data for deliveries and system users.
- Security system implementation - using OpenCV and Python Flask to set up facial recognition features via video within the mobile app as well as on the robot camera and communicate that information to a central server.
| Item Name | Type | No. of Units | Per Unit Cost (in Rs) | Total (in Rs) |
|---|---|---|---|---|
| Total in (Rs) | 40041 | |||
| Raspberry Pi 3B+ | Equipment | 1 | 6375 | 6375 |
| 70:1 Metal Gearmotor 37Dx54L mm 12V with 64 CPR Encoder | Equipment | 2 | 3693 | 7386 |
| Arduino Mega | Equipment | 1 | 1450 | 1450 |
| 5 inch LCD screen for Raspberry Pi | Equipment | 1 | 5000 | 5000 |
| Raspberry Pi Camera v2 | Equipment | 1 | 4500 | 4500 |
| 10X10 Inches Plastic Acrylic Sheet | Equipment | 2 | 270 | 540 |
| LiPo Battery | Equipment | 1 | 3850 | 3850 |
| SD Card with Reader | Equipment | 1 | 2300 | 2300 |
| Caster Wheel | Equipment | 2 | 170 | 340 |
| Pololu Dual VNH5019 Motor Driver Shield for Arduino | Equipment | 1 | 6300 | 6300 |
| Connectors (Screw Terminals, Barrel Jacks, Pin Header, Bus Connectors) | Equipment | 1 | 1000 | 1000 |
| Vero Boards and Soldering Equipment | Equipment | 1 | 500 | 500 |
| Printing, Stationery and Oveheads | Miscellaneous | 1 | 500 | 500 |